what does this mean? "This is precisely because a library should not be deterministically recompiled for all users of the library." - rust

I am new to Rust and is trying to understand the Cargo thing. I read in their FAQ about "why do binaries have Cargo.lock in version control, but not libraries?" but do not understand what this means.
"Users dependent on the library will not inspect the library’s Cargo.lock (even if it exists). This is precisely because a library should not be deterministically recompiled for all users of the library.
If a library ends up being used transitively by several dependencies, it’s likely that just a single copy of the library is desired (based on semver compatibility). If Cargo used all of the dependencies' Cargo.lock files, then multiple copies of the library could be used, and perhaps even a version conflict."
Appreciate if anyone can explain this. Thanks.

Say we have the following crates:
string-tools: this is some kind of commonly used library exporting a FastString struct that has faster implementations of some commonly needed functions. This library has two versions: 1.0.1 and 1.0.2. Version 1.0.2 was recently released.
database: a library to interface with your favorite database. It needs to do some string processing, and so it uses the string-tools library. One of the public methods in database has a signature like this:
fn get_username(id: u64) -> string_tools::FastString
The library has not had a chance or need to update to version 1.0.2 of string-tools - maybe it's unaffected by any of the bugs that were fixed in the patch. Consequently, it's Cargo.lock pins the version of string-tools to 1.0.1.
client: this library is for interacting with the client. It also depends on string-tools, and has a method like this:
fn show_name(name: string_tools::FastString)
This library is using the most recent version of string-tools, version 1.0.2. That is also the version in its Cargo.lock.
You would like to write a website that makes use of the database and client libraries. When you build your project, Cargo compiles each library's dependencies with the versions that are specified in the Cargo.lock. This means that database uses version 1.0.1 of string-tools and client uses version 1.0.2. But now imagine that you've written some code like this:
client::show_name(database::get_username(id));
This should compile. After all, the get_username function returns a string_tools::FastString, which is accepted by the show_name function. But it doesn't! That's because the FastString returned by get_username comes from version 1.0.1 of string_tools, while show_name wants a FastString from version 1.0.2! This can't work; after all, when the patch to string-tools for version 1.0.2 was written, it's possible that the author added an additional field to the type. There's nothing reasonable for the compiler to do.
This kind of issue is avoided by having Cargo ignore the Cargo.lock file on libraries. Instead what it does is compile both database and client against the 1.0.2 version of string_tools. This is correct because 1.0.1 and 1.0.2 are "semver compatible versions." That means it is ok to change out one for the other, things must still compile. Now you no longer get a compiler error, because the types that the one function returns and the other function accepts are no longer different.

Related

Is it possible to specify version for feature in dependency in Cargo.toml?

For example, I use barcoders crate:
barcoders = {version = "0.10.0", features = ["image",]}
Is it possible to specify which version of image this dependency should use?
Something like
barcoders = {version = "0.10.0", features = ["image=0.22.3",]}
Because it uses image crate version 0.18.0 and in my project I use latest 0.22.3.
Does it mean that there's only 2 ways to resolve that:
I downgrade version in my package
Barcoders dependency get updated
No, there is no way to specify the version for a dependency's (optional) dependency. This makes sense, as your dependency run their tests only against the version they specify in their Cargo.toml. In this case, as it appears everything you're doing uses open source, you could fork barcoders, update the dependency, run the test suite and if it passes, use your fork. It would also be polite to open an issue in that case.
If barcoders wasn't open-source, so you couldn't fork it, your best bet would be to switch to the version of image that barcoders uses. If your crate is a library, it may be annoying to expose a public interface that uses outdated libraries, but that's life. The "proper" solution to this problem is to wait until image has a 1.0 release, which is basically a forward compatibility promise, then barcoders can specify image = "^1" (i.e. >=1.0.0 <2.0.0). I mention this "solution" only because you appear to have commit privileges on barcoders, in fact you solved your own problem by updating the image dependency in barcoders.
As one of the comments points out, this version compatibility issue is less fragile that it at first seems. So long as types from different versions of some dependency crate don't cross api boundaries, your project can include any number of versions of that dependency simultaneously. Working with multiple versions of libraries took some work from the rust team on name mangling, which you can read about here
No, you can't, and shouldn't, and shouldn't worry.
Libraries were developed at a single point in time, used dependencies with a certain API. The dependency is likely to change some of that between major versions (changing the type a function returns, exposing different patterns, or whatever). This may make it unable to compile anymore. To really update something, you might need to change parts of the code that is using the dependency in the first place.
This is open source world, so you can do so and publish a pull request in the original crate to update. It might be appreciated, but don't underestimate the care that needs to be taken to not break other people's crates yourself when doing so.
Or make your own fork of the crate that updates it just for you.
But you are probably just worried seeing duplicates of the same crate with different versions during compilation. Cargo indeed compiles with different versions, so all calls to the dependended crate will receive what the developer expected when he/she wrote it. This is not a problem, in performance, or amount of instructions that end up in the binary. Just stop worrying.

Is it documented that Cargo can download and bundle multiple versions of the same crate?

By forking and playing with some code, I noticed that Cargo can download and bundle multiple versions of the same crate in the same project (native-tls 0.1.5 and 0.2.1, for example). I have wasted so much time by looking at the documentation of the wrong version.
I have looked for some information about this behaviour but I was not able to find anything. Is this documented somewhere?
Is there an easy way to determine/detect the version used by the code you're working on (current edited file)? Or can we tell Cargo to show some warnings/prevent the build if two versions the same crate are required?
It was a conscious decision when designing Rust to allow multiple versions of the same crate.
You have probably heard of Dependency Hell before, which occurs when 2 (or more) dependencies A and B have a common dependency C but each require a version which is incompatible with the other.
Rust was designed to ensure that this would not be an issue.
In general, cargo will attempt to find a common version which satisfies all requirements. As long as crate authors use SemVer correctly, and the requirements give enough leeway, a single version of the dependency can be computed and used successfully.
Sometimes, however, multiple versions of the same dependency are necessary, such as in your case since 0.1.x and 0.2.x are considered two different major versions. In this case, Rust has two features to allow the two versions to be used within the same binary:
a unique hash per version is appended to each symbol.
the type system considers the same type Foo from two versions of C to be different types.
There is a limitation, of course. If a function of A returns an instance of C::Foo and you attempt to pass it to a function of B, the compiler will refuse it (it considers the two types to be different). This is an intractable problem1.
Anytime the dependency on C is internal, or the use of C is isolated, it otherwise works automatically. As your experience shows, it is so seamless that the user may not even realize it is happening.
1 See the dtolnay trick that crate authors can use to allow some types to be interchangeable.
Cargo can indeed link multiple versions of some crate, but only one of those versions can be a direct dependency. The others are indirect references.
The direct reference is always the version referenced by Cargo.toml and on top-level of Cargo.lock (while the indirect references are in the dependencies subsections).
I am not sure how much it is documented, unfortunately.

Is it possible to run cargo install with a specific date instead of version numbers?

I want to install a package and all its dependencies as they were at a specific date and time in the past.
I need to use a slightly older version of rustc-nightly, and therefore I need to make sure that all dependencies pulled by cargo install compile against that old version of the compiler.
Currently, when I specify the version of the top-level package to install, it still seems to pull the latest version of some dependencies, which don't build with the old compiler.
No, this is not possible.
Your best options are:
Upgrade the compiler. If you "can't" do this, evaluate why you can't and decide how much benefit you are getting from that.
Add the dependencies to your own Cargo.toml pinned to an older version that does work.
You can try forking the crate index and rolling it back, but there's no guarantee that will work either.
seems to pull the latest version of some dependencies
Yes, most libraries specify dependencies with a semver-compatible range, such as my-library = "1.0". This will allow any version from 1.0.0 to 1.x.y.
Unfortunately, there's no consensus on whether requiring a new version of Rust constitutes a semver-breaking change yet.
See also:
What exactly is considered a breaking change to a library crate?
How to generate Cargo.lock based on last month's crates.io?

How can I use a preprocessor in Rust to detect the version number? [duplicate]

In C++, you could use something like __clang_version__. Is there something similar for Rust? I searched on the internet, but found nothing.
Not directly.
There is the rustc_version crate which tells you the version of rustc accessible on the command-line; this is designed to be used in a build script. There's also rustc_version_runtime which does something similar, but exposes the information as a runtime call (i.e. it detects the compiler version at compile time, but exposes it at runtime).
Standard disclaimer: be very careful writing anything that depends on compiler version. You should ideally only test for minimum versions for which features are supported using semver (which both of the above libraries support directly).

Is there a way to detect the compiler version from within a Rust program?

In C++, you could use something like __clang_version__. Is there something similar for Rust? I searched on the internet, but found nothing.
Not directly.
There is the rustc_version crate which tells you the version of rustc accessible on the command-line; this is designed to be used in a build script. There's also rustc_version_runtime which does something similar, but exposes the information as a runtime call (i.e. it detects the compiler version at compile time, but exposes it at runtime).
Standard disclaimer: be very careful writing anything that depends on compiler version. You should ideally only test for minimum versions for which features are supported using semver (which both of the above libraries support directly).

Resources